After having done lots of QSOs with the 7MHZ QRO transceiver I found that the receiver still had to be improved. The sensitivity was great, the sound also was but there were some difficulties when operating the radio during evening and night times because some (minor) interference was audible. This symptom had been caused, as usual, by strong broadcast stations transmitting from 7200+ kHz 41m-band. Occasionally “Radio China International” and “Radio Romania International” were discernable. But cure was on the way: The transceiver has a modular concept. Based on this I decided to do a full reconstruction of the receiver module.
The mixer, which is the most crucial part in a 7MHz receiver was changed to an IC mixer using an SL6440 double balanced mixer (formerly produced by Plessey).
The mixer IC offers a very good IMD3 performance (30dBm maximum according to datasheet) plus some decibels (1 to say exactly!) of gain and thus is a good alternative to the dual-gate MOSFET I had used before.
As a special feature there is an input (pin 11) where a current can be applied that determines the overall mixer current. The higher this value is set, the better the IMD3 performance will be. Max. power dissipation for the IC is 1.2 watts but that will require a heatsink. I found that a resistor of 820Ω will lead to a current of 4mA (13V VDD) on pin 11 line and produces good performance without thermally stressing the IC. For optimized IMD3 performance the SL6440 ic should be run in balanced mode.
The receiver schematic in full:
From the left we start with a two pole band filter for 7 MHz. LC coupling again is very loose what reduces the receiver’s tendency to overload.
Next is the SL6440 mixer ic. Input and output are equipped with broadband transformers (data see schematic, please!). The purpose is to convert an unbalanced signal to a balanced one and vice versa. According to the respective entry in data sheet running the mixer in balanced mode enhances performance. Pin11 is used to control the DC operating conditions of the mixer, a resistor (820Ω) sets appropriate bias for mixer stage. The 3 diodes (1N4148) supply correct voltage to a pin that is called “VCC2” which should be slightly lower than VCC supplied to the output stage. 3 diodes in series produce the required voltage drop.
In experiments it has turned out the even when gain of the mixer ic is only about 1 dB the resulting output of the whole receiver is higher than that of its predecessor and taking into account that receiver generated noise is not a problem on the lower short wave bands, there is no rf preamplifier.
If you encounter birdies maybe the signal level of the VFO is too high. Then switching a smaller capacitor into the VFO signal feedline is the best idea.
Next stage is the filter switch that has been copied from the previous schematic.
This stage contains the well-known MC1350 by Motorola. To simplify this section a minimum design has been chosen, Output is unbalanced and broadband. Input also. The only filtering in the whole interfrequency section is done by the SSB filter prior to the interfrequency amplifier. A 100uF capacitor in VDD line helps to suppress audio frequency feedback and self-oscillation in the receiver strip.
As you might have realized the transceiver not longer is a “NE602 free zone”, because this mixer now serves as a product detector. A type of usage where the low IMD3 performance does not matter. The low pass filter by the end of the mixer must be chosen according to the user’s preferences concerning pitch and tone.
Audio amp section
Audio preamp is again an ic, the “antique” LM741. Negative feedback has been set to an amount that there is significant gain in this stage (R=330kΩ).
The audio final amp here has been equipped with the TBA820M integrated circuit, the smaller version of the 16 pin TBA820 integrated audio amplifier. The advantage of this ic compared to LM386 is lower distortion and the fact that this ic is not so prone to self-oscillate.
Automatic gain control nearly is the same like in the former version. The main difference is that MC1350 needs positive voltage to reduce amplifier gain. Thus the output has been punt into the emitter line. The problem when using an NPN transistor in such a circuit is that maximum voltage is limited to Vmax = VDD – VBE. As a consequence you can not get full 12V out when you supply 12V between C and E. Here this does not matter because AGC significantly reduces gain already when reaching 6 volts (Source: Datasheet):
Maximum gain reduction (>60dB) occurs between 6.7 and 7 volts.
Also a manual method to reduce gain has been applied. This is by chosing a voltage between 0 and 12V using a potentiometer. To prevent current flowing from the center of the potentiometer into the ADC input detecting the AGC voltage a silicon diode has been installed. To prevent a short circuit of the AGC voltage against GND when the potentiometer is at 0 position (delivering full gain in the MC1350 amp ic) the 5.6k resistor is used.
Changing time constant can be achieved by a second capacitor set in parallel (either by a switch or by microcontroller).
To protect the analog-digital-converter (ADC) in the microcontroller from excessive input voltage, this is limited to 5.1V by a zener diode.
The receiver is very sensitive. Reception is possible with the famous “wet finger” ;-). With a large antenna (full sized delta loop) no overload is detectable even during evening and night times. Noise is slightly higher compared to that with the MOSFET equipped receiver but very much acceptable for a 7MHz receiver.
Hi again! This project directly “beams” you back to the “Good ol’ 80s” when there was no stuff like “DDS, “OLED” or even “SDR” or other modern technology we today use to build our radios.
I designed this transceiver using the “old school” techniques because in a German QRP forum on the internet some hams originated a “Back to the roots”-movement which I thought was a great idea. So I too went back in time 3 decades and constructed a radio like I did it in the eighties at the beginning of my “homebrew career”. That meant: No digital stuff, just a simple VFO but (and that is new) higher rf output power because condx are fairly low on the hf bands currently.
I later presented this radio at an annual German convention of homebrewing hams called the “Black Forest Meeting” named by the place where it is held the beginning of October each year.
To give you an impression, that’s how the radio looks from the outside. Pretty “old school”, isn’t it?
The main design objectives were very simple:
Compact in size (even without using SMD components),
Analog VFO with vernier drive (1:10 gear) and variable capacitor,
No digital stuff (=> no digital noise!),
RF Output in the range from 15 to 20 watts pep in SSB,
Single conversion superhet (9MHz interfrequency)
No “save as many components as possible”-design.
First the block diagram giving you the basic structure of this radio::
“Old school” SSB transceiver for 14MHz by DK7IH (2018) – basic outline
I decided to use an analog VFO in this project due to three reasons:
It’s really old style,
it is much less prone to produce any unwanted “birdies”, and
phase noise performance usually is better than most of the digital ways to generate a signal.
For the VFO I chose the Hartley design characterized by a tapped coil. This type uses less critical components than a comparable Colpitts circuit thus reducing number of parts (2 caps in this case that are avoided) which might lead to unwanted frequency changes (drift).
How to build a VFO that is really stable
Lots of pages have been written about this topic. This another one. First, be aware of the fact that it is not possible to build a VFO that has the same frequency stability like a modern digital system. This is because these systems are all crystal controlled. But it is possible to achieve a drift of some dozen Hertz within an hour or so which is absolutely sufficient for having even a longer QSO.
The main problem is based on physics, or thermodynamics to say more exactly. All material expands when heated and contracts when ambient temperature decreases. OK, some exceptions exist, water below 4°C is the commonly known example of them.
Avoiding thermal runaway
Heat is the problem in such a circuit. It comes from the interior of the components when current flows through them and from the outside, for example when the transceiver is exposed to sunlight or placed near another source of thermal energy. Also heating of the final rf amplifier stages may contribute to heating the cabinet inside. The electronic parts forming the central strucure of the tuned circuit exert the main influence connected to thermal runaway of the frequency that is generated.
The general approach is: When we can’t avoid physical effects we must choose components that change their values in such a way to compensate the changing of the values of the other parts. That means we have to look carefully on temperature coefficients of the varoius components we intend to build into our VFO.
Choosing the “right” components for your VFO
Choosing advantageous components is crucial for frequency stability. So I did some brief research to find out more about temperature coefficients of coils of various types and available capacitors. Here are some of the outcomes.
Explanation of syntax: If a relation is negative, a minus sign (“-“) is given. In this case the value (C or L) decreases when temperaure increases. A plus sign (“+”) indicates a positive coefficient. When the relation of value change by temperature change is weak (that means no intense changing of the value when heated), there is only one “-“-sign. The more “-“-signs you have, the higher this respective ratio is. The same applies for “+”-signs to indicate a positive relation.
Ceramic capacitors: —
Polystyrene capacitor: –
NP0 (C0G) capacitor: no measurable effect
Air coil on polystyrene coil former: +++
Coil wound on T50-6 yellow toroid: +
Based on this short survey, the best combination would be NP0- and Polystyrene caps together with an inductor wound on an T50-6 (yellow) core. Hopefully their temperature behavoiur will compensate more or less and lead to best stability. Hint: On the photos appearing later in this text you will see an air coil wound on a TOKO style coil former that has been used because it does not need so much space.
The VFO circuit
I finally chose the Hartley circuit for my VFO. There it is:
VFO Circuit explanation
Starting from the left you can see the tapped coil (here 60 turns tapped at 10 turns from the bottom end) on a 5.5 mm TOKO style coil former without any core. In parallel there are various capacitors (polystyrene and NP0 mixed) to build up the total capcity. It is common use to spread the total capacity needed to various single capacitors because it has turned out that the effects of temperature change are less significant if you use more (and therefore smaller) single capacitors.
A 100k resistor is used to pull the gate to ground and therefore provides a correct bias at the FET’s gate. The 1N914 diode is a so called “clamp” diode that has been installed to stabilize (and therefore reduce) the rf voltage in order to avoid excessive rf voltage coming to the FET’S gate which would lead to distortion. This diode has a negative side effect, but that an be accepeted for a VFO in the rf bands: It slightly increases phase noise because it works as a regulator. With some the designs you can see this diode in reverse position, don’t worry, the regulating effect takes place either.
To ensure the oscillator to produce radio waves, in-phase feedback between gate and source is generated via the tap you can see with the coil. A tap of about 1/6 of the whole number of windings provides enough feedback voltage to let the oscillator start by its inherent thermal noise and generate clear sine waves afterwards. Putting the tap too close to the “hot end” will cause distortion because the amount of energy coupled back to the gate will be too high. Also instabilites are probable because of excessive drive power to the gate of the FET.
On top of the tuned circuit there is a varactor diode that is used to be controlled by a positive voltage to form a RIT (receiver independend tuning) control circuit. It is very loosely coupled to the tuned circuit to minimze temperature effects and because only 1 or 2 kHz “swing” is needed. The generation of the RIT voltage will be described later in this text.
The main tuning capacitor
An air capacitor is mandatory here! You can either use a ready made one from the surplus market. But to keep it as small as possible I built my own by dismanteling one of old variable capacitors formerly used for homemade AM radios. Use a small drill to remove the rivets, dismantle the capacitor completely and rebuild it again as an air capacitor (get rid of the plastic dielectric interlayers!) by using M2-screws and nuts. Youl will have evenings of endless fun with this game! 😉
Buffering and amplifying the Signal
The second stage with another FET is very loosely coupled to the source of the first FET. This is made to minimize effects of load changes to the frequency. This stage is a so called “source follower” giving a very low impedance signal to the final stage that is responsible for the amplification of the signal to a level of 2 to 3 volts pp which you will need for the rx mixer that has been designed as a dual-gate-MOSFET mixer (see receiver chapter later!).
DC voltage in the VFO
Voltage stabilization is crucial for best performance of this critical part of the radio. Supply voltage changes always lead to frequency changes. So a two-level buffering is common use here. The first (and most critical) stage is buffered twice (10V voltage regulator integrated circuit 78L10 and subsequently by a 6.2V zener diode) whereas the buffer and the amplifier stages are supplied with 10V regulated DC voltage only.
Ambient thermal isolation
To avoid the VFO being affected by interior thermal convection (flow of warm air inside the cabinet) it is recommended to shield the VFO from the rest of the transceiver. I do not recommend using metal sheets as walls here because these form other unwanted capacities that will lead to thermal effects on the generated frequency. Metal also is a good conductor for thermal energy, so you might run counter to your goals. My thermal insulation therefore is made of simple cardboard.
The Local Oscillator (LO)
This oscillator is much more uncritical than the VFO because it is crystal controlled. The purpose of the LO is to supply a carrier signal for the SSB modulator. Due to the fact that there are two sidebands we theoretically can use this LO must be switched to either one of two possible frequencies. In case of an interfrequency of 9MHz (9000kHz) these are: 9001.5 kHz for the first sideband and 8998.5 kHz for the second sideband. Please note that I did not write “USB” or “LSB” because the frequencies forming each sideband might be changed because of the frequency plan of the transceiver where by mixing with the VFO frequency the sideband might be inverted depending on if you add or substract the VFO frequency from the 9MHz-SSB signal.
There are several possibilities to produce these two frequencies:
Using two different oscillators each equipped with a single crystal,
switching two crystals with one osillator,
using a variabale capacitor or a coil to “pull” one crystal to the desiered frequency.
This method means high effort but surely is the most exact one because there are no influences of the unneeded choice to the other crystal currently on duty.
Is the worst idea because the unswitched crystal is highly prone to influence the freqeuncy of the switched one because they are linked to parasetic capacities within the wiring, the switch and so on. Forget this one espacially when using the internal oscillator of a NE602/SA612!
This to my point of view is the best compromise between circuit simplicity and function. You can see this way of sideband switching in my transceiver.
This is my local oscillator:
It is a simple Colpitts circuit where in-phase feedback and feedback voltage control are achieved by a series of two identical capacitors. A simple switch, a capacitor (90pF max.) and a coil (4 to 8uH max.) that are either connected to the base of the transistor via the 9MHz crystal determine the sidband freqeuncy of the oscillator. Signal is taken out via the collector.
The receiver will be presented step-by-step starting with the front end stage:
The RF Preamp
This stage is connected to the antenna relay. It provides an amount of basic amplification for the antenna signals. But that is not the main purpose. Noise figure improves significantly if you use a stage with low inherent noise. Thus a dual-.gate MOSFET is installed here. This semiconductor is also used to control stage gain because gate 2 of the MOSFET is connected to the AGC chain of the transceiver. About 12 dB gain swing are possible here. Stage gain is about 15dB.
Note the position of the primary and secondary windings of the input and the output coil. To avoid self-oscillation the output (drain) of the MOSFET is connected to the untuned part of the LC circuit. Coils must be shielded and should be equipped with ferrite heads (in the photo the piece in left bottom corner).
The Receiver’s Mixer
In this stage also a dual-gate MOSFET is used. This type of mixers provides good capabilities to cope with high signal levels without producing unwanted signals (high IMD3), gives some dB of gain and is low-noise also.
One “disadvantage”, if you want to say so, is that it needs a little bit of higher VFO drive (about 2 to 3 volts pp). Gate 2 bias is generated via the voltage drop on the source line. The tuned circuit in drain line is adjusted to the desired interfrequency. See the schematic for the exact winding data and parallel capacitor.
The SSB Filter section
Transmitter and receiver share the same SSB filter in this transceiver. So some sort of switching is recommended even if circuits exist that go without one. I used a high quality relay made by Teledyne that I bought in a 10-piece bundle for low price (1€ each!) via a well-known internet marketplace. Caution: Some SMD-relay I tested prior to building this rig were disastrous concerning signal isolation between terminals. To avoid any disappointment or frustration I recommend testing a relay before you finally install it.
All connections to the rest of the circuit must be made with shielded cable. I found an interesting alternative: I sometimes design my own very thin shielded “cable” with brass tubing (1 mm inside diameter) where I put insulated cable inside. The brass tube is connected to GND on the Veroboard. You can not bend these tubes but longer lines can be interrupted for a short piece so that the “bend” can be made by putting two parts of tubing in 90° degree angle for example.
The IF Amplifier
This one might look familiar to you. It is a simple “remake” of the front-end stage. The one remarkable thing is the secondary of the output transformer. This coil has 4 windings (prim. 16 turns). The secondary is center tapped (2 + 2 turns). This is because the product detector (SSB demodulator) has a symmetric input. Very important in this stage is the 100uF capacitor in VDD line. This cap prevents the stage from AF resonating and self-oscillations on the VDD line and makes the receiver much more “quiet”.
The SSB demodulator
This stage is probably the most “old-school” part of the whole transceiver. It uses an old CA3028A differential amplifier as mixer circuit:
You won’t be able to buy large amounts of this IC anymore. And if you get one, the prices are close to or beyond a rip-off. But there is an alternative. You can build your own “IC”. Watch this page where all the information you need is provided!
Signal input goes to the paired transistors forming the amplifier stages. LO is fed into the line transistor that is used to set the current of the differential amplifier thus providing a switching and therefore superposition of the two signals.
The output circuit is made of an audio transformer formerly used in the final audio amp of old AM radios (coil resistance is about 300Ω each side). The 2.2nF capacitors eliminate remainders of the rf signals and “ground” the terminals of the AF transformer.
This final receiver part consists of two stages: An audio preamplifier with a bipolar transistor and a final amp with a TBA820M integrated circuit.
The two caps 0.22 and 0.1uF determine how the higher frequency components in the audio signal are cut off. The higher the total value the more the higher frequency components of the audio signal will be limited due to the equation XC=1/(2*PI*f*C).
Tr1, which is a universal purpose NPN transistor, provides high gain. Thus a 10k resistor is installed to form a voltage divider with the audio gain potentiometer.
In the final stage I use a TBA820M ic (8-pin DIL version). This one is more linear than the well-known LM386 that you usually can find in this place and it is not so prone to self-oscillation. The cap aside the 100uF in the top left corner of the schematic is not marked, its value is 0.1uF.
Loudspeaker impedance is 8Ω.
Automatic gain control makes listening to signals much more comfortable. AGC voltage is audio derived, like in my other transceivers. The circuit also is nearly the same:
Due to the very high gain of the product detector this stage is directly connected to this circuit and not to the AF preamp. A potentiometer is used to set the threshold of the AGC onset.
Next stage is a simple audio preamplifier followed by a “Greinacher Circuit” serving as voltage doubler and providing DC voltage proportional to the audio signal level. A dc amplifier with another NPN transistor lets its collector voltage drop as soon as it is fed with significant dc input voltage. Thus this voltage decreases and so it can be used to control gate 2 of the MOSFETs in the various receiver stages that are equipped with tetrodes.
The S-Meter is connected to the emmitter of the final transistor. If conductivity in the transistor rises, the emmitter becomes more positive and the S-Meter needle is deflected proportionally. The 220Ω potentiometer in the emmitter line must be set in accordance to the respective S-Meter you are using. One shortcoming should be mentioned: If you have a not so sensitive meter then the value of the pot can be set to nearly 100Ω or above. This will prevent the collector from dropping to nearly 0V in case there is a strong signal and hence reduce the maximum dB you can get from the AGC chain.
The transmitter section is designed for an output level of about 20 watts and uses 4 stages all equipped with bipolar transistors. The last stage is a push-pull stage, the 3 low-power stages are single ended. I prefer push-pull for the last stage (if possible) because this circuit inherently does not create even harmonics thus simplifying output filtering.
The first parts of the transmitter to be shown here are the microphone amplifier, the SSB-generator and the TX mixer:
The mic amp is simple but provides enough gain and good linearity for using an old-style dynamic microphone. It works in common emmitter mode and has gain of about 15 to 20 dB.
The audio signal amplified by the microphone amplifier is fed into PIN1 (Input 1) of an NE602/SA612 mixer IC which is the simpliest way to generate a DSB signal with a Gilbert cell. LO input is fed to PIN7 and should be in the range of 200 to 300mVpp. Thus a 12pF cap has been installed to limit LO voltage going to input at PIN7.
Carrier suppression is around 45dB when LO offset frequency is correctly set for each of the two sidebands and LO voltage is not much higher than the 300mVpp mentioned before.
The DSB signal produced by this mixer goes on to the SSB filter relay and filter that has been described before. The use of shielded cable is mandatory, too.
The TX Mixer
You won’t be able to recognize many differences if you compare this TX mixer to the DSB generator. In fact, there are none.
The 14MHz Band Pass Filter
Next is the band pass filter that consists of 2 coupled tuned LC circuits for 14MHz. They are also wound on TOKO style coil formers. Data can be found in the schematic underneath.
It is important to also install the ferrite heads that are provided with most of the coil formers and to use the shield “metal cans” that are also standard for these coils. This is to prevent stray coupling of rf energy into the first stage of the power amplifier strip and therefore preventing self-oscillation of the transmitter strip.
For proper adjustment set the transmit frequency to about half of the frequency swing ((i. e. to about 14.200 kHz) and tune for max. output.
If you modulate with a two-tone signal to the mic amp you should see about 500mVpp by the output of the BPF when the chain is fully driven.
We start with the low power end of the power transmitter section. A bipolar rf type transistor is the center part of this stage.
This one is a standard circuit and has been “trimmed” for maximum linearity in order to reduce distortion to a minimum (which is also true for the following stages). You can see the well understood 2 master ways of achieving max. linearity in an amplifier stage:
Negative feedback between collector and base (i)
Emmitter degeneration (II)
i) The first measure goes along with the 2.7kΩ resistor between collector and base of the transistor. This resistor provides positive dc bias voltage to the base and leads 90° out-of-phase ac voltage to the transistor’s input. This reduces gain and therefore distortion. But due to the fact that the whole transmitter strip has plenty of gain, this loss in gain is not a serious problem.
ii) The 10Ω resistor in the emmitter line is not bypassed by a capacitor. This stabilizes the circuit. When the current through transistor increases the emmitter voltage will rise (according to Ohm’s law) and the voltage between collector and emmitter drops. This reduces voltage difference between base and emmitter and hence also reduces gain.
The coupling to the next stage is done by a capacitor of 0.1uF. This causes some impedance mismatch. But that is as well not a big problem because the gain reduction here helps to prevent the whole transmitter from unwanted oscillations by diminishing overall gain.
This stage is somehow a copy of the stage before but allows more current to flow through the stage. It is also operated in class “A” mode and uses the same methods to maximize linearity like the preamp stage.
You can use a 2N3866 transistor here which is available. But any other rf power transistor for driver stages (2SC1973 etc.) will also do the job well. A heatsink is recommended even if stage current ist not that high. T1 should be a toroid, a “pig-nose” core in this place to my experience is not the best choice. The 10uH RFCs are ready made ones but you can also wind 20 turns of 0.4mm enameled wire to a FT37-43 toroid core.
RF output of this stage could be measured as 100mW into a 50Ω load.
The Main Driver
This stage has an old 2SC2078 CB transistor and is operated in class “AB” mode. An alternative could be a 2N3553 that is available on ebay for example. A heatsink is neccessary for whatever type you use.
Correctly set bias for “AB” operation is ensured by the 1kΩ resistor from VDD to the bias circuit. The 1kΩ resistor limits the current whereas the diode works as a stabilizing element (thermistor). It must be connected directly to the case of the transistor ensuring good thermal contact. If the temperature of the devices rises the resistance of the diode will decrease. Hence current through the diode increases thus reducing the part of the current that can pass through the base-emmiter line of the 2SC2078. Quiescent current is stabilized and thermal runaway is prevented.
The rf output is uncommonly terminated with a low-pass-filter. This is because I first intended to build the transceiver for an output level of about 4 watts. But then I had the idea that the space still available on the veroboard could be used by another amplifier definitely leaving the QRP power level. So I left the circuit how it first was and just added the final amplifier stage.
Output of this driver stage now ist set to 1 watt into a 50Ω resistor.
The Final RF Amplifier
Now let’s go for the power machine in this transceiver:
2 rf power transistors 2SC1969 by Eleflow provide up to 20 watts of rf power. Bias for such a high power stage can not be set by a simple resistor. Here a line transistor (BD137) serves as current control. Diodes D1 and D2 (1N4002 or equ.) follow the same purpose like the single one in the stage described before. They must be mounted with excellent thermal contact to each of the 2 power devices which ensures secured protection against thermal runaway. The transistors also must be connected to a large heatsink. I use Aluminium metal strips (2mm thickness) to connect them to the back wall of the cabinet.
RF is fed into the power transistors via a network of 8.2Ω resistors and two 22uH rf chokes that seperate the rf line from the dc bias line letting only dc pass. This method makes construction of the input transformer easier. Winding ratio is 4 turns primary, 2 turns secondary. This is because the input impedance of the stage ist fairly low (aorund some ohms).
The output transformer is a homemade “pig-nose” of 6 toroids FT50-43, where 3 toroids are stacked (using 2-component glue) and 2 of these stacks are glued in parallel (see picture at the end of this text for details!). Winding ratio is 1 + 1 (primary center tapped) to 4 on secondary.
Quiescent current of this stage should be set to about 100mA.
A low-pass-filter terminates this stage and is connected to the antenna relay.
In addition you find a section to measure rf power. This is again the so called “Greinacher-Circuit” which doubles the voltage and serves as a charge pump. The dc output of this circuit directly leads to the S-Meter indicating output power of the transmitter.
First the spectrum of the signal with the transmitter fully driven to 20 watts output power with a two-tone-signal:
IMD3 is about 28dB below signal peak which I think is acceptable.
Amplitude diagram is as follows:
Max. radio frequency voltage is 90.4Vpp which calculates to about 20 watts of rf power (P=(Vpp/(2*SQR(2)))²/50Ω).
Power switch board and RIT voltage
A 12V relay with two pairs of contact sets is the heart of this unit. DC power is lead to TX, RX and permanent supply via the respective power lines.
RIT voltage generation is a little bit more complicated. When the RIT switch is in “OFF” position, RIT voltage always is taken from the fixed voltage divider that is formed of the two 4.7k resistors either when on receive or transmit mode.
If RIT is “ON” then there are two possibilities: When on receive mode, RIT voltage is gained from the 10k lin. potentiometer in the front panel. When on transmit mode RIT again is taken from the fixed voltage divider.
There is also a false polarity protection diode. This can be any silicon type with max. current >= 5 A.
The construction is sandwich style made of 2 layers:
OK, that’s the story. Thanks for joining me on the trip to the past! 😉
The challenge started some weeks ago, when John, ZL2TCA, commented to this blog
you next challange is to build a rig into a cigerette packet size case.
My problem: I don’t smoke, have never smoked and probably never will. 😉 But I have a new transceiver for 20 meters, that might come close to the dimensions of a pack of “cancer sticks”.
The transceiver is nearly the same circuit as applied with the “Micro 20-III” but uses a single ended final amplifier instead of a push-pull circuit. I hope to find time the next days to publish an article on this rig featuring full description of the radio. Currently I’m in the IOTA contest and working stations from all over Europe.
Having deferred the work on the “micro multibander” for some time I finished another small QRP rig (this one for 7MHz) that is suitable for my summer excursions by bike or hiking the local mountains here in the State of Rhineland-Palatinate or the Black Forest that is not that far away on the other side of the Rhine valley.
Besides, this transceiver to be discussed here is some sort of a “remake” of a 20 meter rig I built 3 years before. And this time, the transceiver really fits into a shirt pocket without having to wear “XXXXL”- clothing. ;-):
General circuit description (instead of presenting a block diagram)
The rig uses two mixers NE602 plus one filter as central elements. The signal way is reversed when switching from receive to trasmit mode. This is done by 2 relays and is a well known technique for simple QRP rigs. You will find lots of equivalent ideas on the internet (Example 1, Example 2).
But not to ignore the shortcomings of these designs: They are somehow inferior to my requirements, particularly concercing receiver performance. I prefer to have higher signal gain and an AGC circuit. AGC for me is a must. But these designs can be expanded easily, so I added an AGC controlled interfrequency amplifier with dual gate MOSFET BF998 into the receiver’s signal path enhancing performance significantly.
The frequency generation of the superhet transceiver scheme is simple: Again I use one interfrequency (i. e. 9MHz). The VFO is DDS based on AD9835 operating below the desired radio frequency, which means that it is set to the range of about 2 MHz. Due to this low frequency you could replace the DDS by a VFO if you don’t like the relatively complex work with the software programming and microcontroller stuff). A 2MHz VFO can also be made very stable, so this is an alternative not to be ignoered.
Due to the fact that the schematic is not very difficult to analyze you are kindly requested to refer to it for further talking:
In the center of the schematic you can see the main elements of the circuit: One SSB filter (9MHz), correctly terminated by 2 resistors of 1k each (to ensure proper filter response curve) and two relays with a double set of switches. These relays reverse the way the signal travels through the filter. The advantage of this: You can use the integrated oscillator of the NE612 controlled by a crystal and a tuning capacitor to set the carrier frequency correctly for the lower sideband because the mixer is used as SSB generator and as product detector in common.
A word on chosing the proper relays: An intense examination of the relays’ data sheet is essential. I built a prototype of this transceiver on a breadboard prior to soldering the components to a veroboard. I found that some SMD relays have signifikant coupling capacities between the unused relay contacts (in the range of some Picofarads). So stray coupling was a severe problem. Later I used some second-hand Teledyne RF relays that I had purchased via ebay two years ago (price originally 50€!) for 1€ each. These relays are absolutely superb!
Before we go: In the circuit scheme above I missed out the antenna switch relay because I think every homebrewer knows what to do in this case. 😉 So the receiver’s signal path starts with a band filter for 7MHz consisting of to tuned LC circuits. The coupling is relatively loose. As coils I use the well known coil formers in TOKO style with 5.5mm outside measure.
Coil data for the 7MHz band pass filter (BPF) is 39 turns primary and 9 turns secondary of 0.1 mm enameled wire. The respective capacitor is 33pF. This is a high L to C ratio which gives you excellent LC quality factor. This is mandatory especially when working on the 40 meter band, because of the strong broadcasters starting from 7.200 kHz intermodulation might be a problem when the receiver is connected to a high gain antenna and broadcasters’ signals might overload the first mixer (remember that NE612 has a relatively low IM3!). If you still should have problems coping with too strong out-of-band signals you can reduce the coupler from 4.7pF down to 2.7pF.
In practical terms I could not detect any unwanted signal products even when using an antenna with high rf output voltage. One reasons for this is, that there is no rf preamplifier for the receiver. This avoids overloading the first mixer generally.
The NE612 has two mixer inputs and two outputs. This makes it very suitable for this sort of radio. In receive mode pin 2 of the right NE612 is used as signal input. VFO signal is fed into pin 6. The resulting mixer products are taken out from pin 4. Next the 9MHz filter follows from right to left.
The 9MHz IF signal then is fed into an IF amplifier. This one is equipped with a dual gate MOSFET (BF998), gain is about 15dB when full AGC voltage is applied wich leads to about 6V by the 1:1 volatge divider in the applied to gate 2 of the MOSFET.
The left NE612 is the product detector. I use the internal oscillator with a 9MHz crsytal and a tuning capacitor here. This saves building an extra oscillator and simplifies the rig again.
One AF low pass filter made of 1k resistor, 100uF rf choke and a 0.1 uF capacitor eliminates high frequency remainders generated by the mixing process.
The audio stages are also made simple: One preamplifier (using bipolar transistor in grounded emmitter circuit) and a final stage with LM386 transform the signal to a level that is sufficient to be fed into a small 8 ohm loudspeaker or a set of standrd MP3-player headphones. Because the rig is very small and there was definetely no space for a loudspeaker I use headphones instead.
Keep an eye on the power supply switching of the two audio stages. The problem was to eliminate the switching click and pops to a minimum and to avoid acoustic feedback when unsing a loudspeaker. So the audio preamp is only connected to DC on receive. When switching to transmit the charged capacitors avoid instant cut off supplying some Milliseconds DC to the amp until significantly discharged. The main amplfier on the other hand is connected to permanent DC supply. So it won’t pop when switching from tx to rx an vice versa but can cause feedback. To avoid feedback a transistor is used to cut the speaker/earphone from the power amplifier.
AGC is audio derived. A two stage amplifier provides a DC voltage analog to the audio input sginal strength. First amplifier stage is a common emitter bipolar transistor supplying sufficient audio voltage. This voltage is rectified by a two diode circuit letting only the positive halfways pass. You can use silicon diodes (1N1418) oder Schottky diodes here. An electrolytic capacitor (100uF/6V) provides the time constant respectively DC decay once the signal has disappeared. Output of the DC stage is split. The collector is connected to 12V via a 4.7k resistors causing a voltage drop when the transitor’s conductivity increases. The emitter is fed to the ADC of the microcontroller (pin ADC1) causing a proportional voltage to the voltage of the applied audio signal so that on the OLED an S-meter can be displayed.
An electret microphone picks the operator’s voice. The signal output level of these microphones is high enough to drive the left NE612 (which serves as balanced modulator in this case) directly. Signal input for the mixer should be 200mV RMS according to data sheet. An electret produces about 0.5 to 1 V pp if spoken with a decent voice in the distance of some centimeters. So you have more than enough audio signal power for the modulator.
BTW: Carrier suppression of the modulator is excellent. I achieved 56dB without doing anything else!
The resulting DSB signal then is fed into the SSB filter, the SSB signal subsequently is directly sent into the right NE612. A band pass filter for 7 MHz eliminates the unwanted mixer products. You should have 400 to 500 mV pp of rf signal here when the transmitter input is fully driven. I recommend a two-tone test generator to check out the linearity of this and the remaining amplifier stages!
Next parts of the transmitter are a band pass filter (same coils and capacitors like th rx bandpass filter), a preamplifier and a driver. The later should put out about 150 mW into a 50 ohm load. They are made more linear by emitter degeneration (4.7 and 2.2 ohm resistors for predriver and driver) and negative feedback. This helps to ensure that transmitter performance is fine when IMD3 products are concerned even if the main IMD3 problems usually occur in the final stage.
To transfer the rf power into the final stage proper impedance matching is mandatory. Input impedance of the final stage is fairly low (<10ohms), therefore a broadband (down)transformer is used. Data is: Core T37-43, primary 12 turns, secondary 4 turns of 0.4 mm enamled wire.
Last stage is a single ended linear amplifier in AB mode equipped with a 2SC1969 rf power transistor by eleflow.com.
BIAS circuit: The combination of the 1k resistor, a silicon diode (1N4002 or equ.) and a capacitor sets up the correct bias. Bias is fed into the cold end of the input transformer. Quiescant current should be around 40mA. A good thermal contact between the diode and the transistor is recommended. As the transistor gets warmer the diode will increase its conductivity so reducing bias current. This will prevent thermal runaway effectively!
To avoid bulky output transformers the PI-filter (7MHz LPF) is part of the tank circuit of the final amplifier transistor. For this power level this is an acceptable and practical solution because the output impedance of the stage is nearly equivalent to 50 Ohms. A certain mismatch is not a severe problem. DC to the final transistor is applied via an rf choke, for exact data please refer to the schematic!
T2 helps to suppress unwanted signals that I encountered when taking the transmitter from the dummy load test environment to a real antenna. I observed unwanted parasetic oscillation in the range of about 1MHz. T2 has a low reactance for this frequency range thus eliminating the oscillations in a reilable way by short circuiting them towards ground.
Powered with 12.5V DC the transmitter will put out slightly more than 5 watts PEP.
AD9835 is a simple but well performing 10-bit DDS chip made by Analog Devices (AD). It is controlled via 3 SPI lines transmitting the frequency data. Maximum output frequency is around 16MHz when the chip is clocked with its maximum clock rate of 50 MHz. Oscillator output voltage is some hundred millivolts peak-to-peak, so you can connect the output directly to pin 6 of the NE612 mixer.
Control signals come from an Arduino Pro Mini board. The microcontroller in this module is, if you are an Arduino user, preinstalled with a bootloader program. I overwrote this small portion of code and use the ATMega168, which is the core of the Arduino, in “native” mode. My software is written in C and transferred via “AVR dude” software using the ISP lines MOSI, MISO, SCK and RESET. These lines are not in the schematic, please refor to ATmega168 data sheet. Alternatively you can use, like shown in the schematic, an ATmega168 controller. So you have to de neccessary wiring on your own.
You will find the source code here. I packed it into an Open Document Text File because of problems I encountered when I tried to store the code into this Blogtext. If you need a compiled HEX-file, please feel free to email me!
Display is a very small OLED with 64 by 32 pixels. The OLED is, to my point of view, a little bit noisy. To suppress any rf traveling on VDD line I use an 82 ohm resistor and a set of bypass capacitors of 100uF and 0.1uF capacity closely connected to the OLED VDD pin to GND.
A low pass filter by the output of the DDS ensures spectral purity and avoids clock oscillator feed through. Remember that if you need another output frequency other than 2 MHz you should redesign the low pass filter.
Tuning is done by a rotary encoder connected to PD5 and PD6 of the microcontroller. I use the pull up resistors internal to the microcontroller, so you won’t see any other things than the mere encoder.
Tunings steps are selected by pushing the encoder knob or another suitable push button. This button is connected to ADC0 in the ATMega168 via a 3.9k resistor. The resulting ADC voltage might be problem because of a certain variation in the values of the pull up resistors that form the second resistor of the voltage divider. There is an outcommented section in the code that will show you the exact ADC value that has to be typed into the code so that key recognition works exactly.
The button once pushed will increase the tuning step by a certain amount of Hz. Steps are 10, 50, 100 (standard step), 250, 500, 1000 and 5000 Hz in and endlessly revolving chain. The step will be reset to 100Hz (standard tuning step) by leaving the tuning knob idle for 2 seconds. That’s all with the controls. Very simple, but sufficient.
The transceiver is constructed on a double sided veroboard with 6 by 8 centimeters area. Components are through hole and SMD where available. The Arduino is mounted to the front panel (another Veroboard carrying the controls etc.) as well as the OLED is. The veroboard is inserted into an aluminium frame connected to the front panel with 4 lateral M2 screws:
Wiring can be made by using the colored lines stripped from old parallel printer cables. These cables have a diameter of precisely 1mm an fit through the holes of the veroboard excactly.
If you connect any external components that are not on the same veroboard use standard 2.54 mm (0.1″) male and female board connectors! This will make it much easier to dismantle and reassemble the rig in case troubleshooting is neccessary.
Use M2 srews instaed of M3 when building very small rigs like this one!
The reverse side of the main arrangement:
Two brass made bends (from the local hardware store and each cut to a length of 8 centimeters) hold the PCB inside the mounting frame. A winding has been cut into the brass to fix the bends with screws in M2.
Together with 2 halves of a bent aluminium cabinet covered with “DC-fix” (a German manufacturer of self-adhesive PVC coating) the final rig looks like that:
DK7IH QRP SSB 40m/7MHz pocket size transceiver – final assembly
So, that’s the end of the story so far. Now it’s time for going outdoor and test the rig in field use. 😉
Work is in progress. The recent weeks I finished all the 6 modules that are going to be the receiver:
Band pass filter section
Relay switches for switching the BPFs
RF preamp, RX mixer and IF preamp
IF main amp
Product detector and AF amp section
Mounted together to an aluminium carrier board it looks like this:
On the picture the board is not equipped with the neccessary wiring yet to give the reader more sight on the single circuits. Next I will draw a schematic of each board to point out the used circuitry for those who want to build this or a similar receiver.
First test are promising so far, the receiver is sensitive, has a very low noise figure (due to dual gate MOSFETs in the preamp and the two main IF amp stages) and has shown no problems to cope with high out-of-band broadcaster signals on the 40 meter band which is due to the SBL-3 mixer I have used that has a good IM3 performance..
After having built my first shirt-pocket transceiver about a year ago I occasionally thought of how this or a more or less modified design could be made simpler to save components and therefore limit space as well as reducing the complexity of the whole rig. This was due to the fact that I thought that the ancestor (see link above!) of this project was somehow „overkill“ because I used plenty of stages redundantly that could have been used for receive and transmit operation.
Before we go into the details of this new project, let’s have a look on the new micro transceiver (here operating portable as EA8/DK7IH/QRP from the island of Fuerteventura):
Cabinet size is about 10 by 4 by 5.5 centimeters which equals to a volume of 220 cubic centimeters (cm³).
Making it more simple without detereorating the performance?
Under the aspect of simplifying the circuit I remembered that I had searched for simple designs of transceiver circuits some years ago intensely. After having revisited some of them on the Internet my attention was caught by the „Antek“-Transceiver that has been published by SP5AHT a while ago. This was an ideal basis for my purpose because I intended to build a „2 mixers+1 filter“ circuit in order to make at least 2 of the 4 mixers that are necessary for a fully functionally SSB transceiver redundant. The central part of SP5AHT’s design matched my requirements in an ideal way:
SP5AHT’s circuit uses one mixer (US2 in the schematic) to serve as the receive mixer during rx periods and for the balanced modulator when on transmit. Then the resulting signal is fed through the filter and subsequently processed by mixer 2 (US3). This mixer works as the product detector on receive mode and as transmit mixer when you are on the air. The two oscillators (VFO and LO) are fed to the respective mixer depending on the current operation. This is done by a simple relay connected to the PTT. So when changing from receive to transmit the two oscillators are swapped thus changing the complete function of the circuit.
To get rid of the relay and because I wanted to use the Si5351A clock oscillator chip, my idea was making two of the 3 oscillators present in the clock chip act as LO and VFO. By software, when switching from rx to tx, these oscillators’ frequencies are simply swapped. The microcontroller driving the Si5351A reads the PTT and when pressed to talk the frequencies present on CLK0 and CLK1 are put out reversely. Thus no hardware switching is required.
In addition the audio amps in the end of the receiver chain are powered off. Instead of this the rf power amp of the transmitter section is connected to +12V DC as well as the microphone amp. The antenna relay disconnects the receiver front end from the antenna line and connects the antenna to the LPF that is installed after the rf power transformer of the rf power amp final stage.
By this the whole transceiver is constructed much simpler and lots of circuitry has been removed from the rig.
The block diagram
Circuit explanation: During receive periods the signal is fed into the antenna jack whose line is switched by the antenna relay and fed to the first band pass filter for 14MHz when listening to the band. Afterwards it is amplified by a dual-gate MOSFET transistor that is connected to the AGC line that reduces gain when strong signals are detected. The next stage is mixer 1 where a VFO signal of about 24 MHz comes from CLK0 of the Si5351A module.
The result of this mixing process is the IF of about 9 or 10.7 MHz or whatever frequency you are about to use depending on the filter you have installed. This signal is amplified by another dual-gate MOSFET. On receive mode this stage is also under AGC control. When transmitting it is powered to full gain of about 18dB applying +12V via a 2:1 voltage divider consisting of 2 resistors with 82k each.
Next step is mixer 2 where the IF signal is mixed with a 9 resp. 10.7 MHz LO signal (receive mode serving a product detector) or with the VFO signal in mixer 2 serving as tx mixer.
On receive two audio stages (a preamp with LPF in advance and a power amp) amplify the audio signal to a level that can be fed into an 8 Ohm loudspeaker.
On transmit a BPF eliminates the unwanted mixing products and a three stage rf power amplifier lifts the signal to a power level of 3 watts peak power.
The Main RF Section
This central section of the rig consists of the two mixers mentioned before, a commercially made (in my case) 10.695 MHz filter stripped from an old CB radio, some amplifier stages and so on. The circuit is the following one:
Starting from left side top there is the first receiver amplifier stage using a dual-gate MOSFET. The use of a MOSFET transistor ensures that the noise figure and sensitivity of the whole receive improve very much. The stage is comnnected to the AGC chain. The dc voltage applied by the AGC section varies in the range from 0 to 6 V DC. In addition it can be set by hand by turning a front panel mounted potentiometer that alters the DC voltage in the range from 0 volts to max. volts from the ADC section.
The input is a single tuned circuit using 4 (antenna side) by 16 turns (MOSFET gate side) on a TOKO 5,5 mm coil former. Parallel capacity is 47 pF. In drain line of the transistor the same filter is used coupling the signal to the first mixer (NE612). Note that the second filter is reversed (secondary in drain line) to avoid self-oscillation of the preamp stage.
This mixer’s signal is fed with 2 different input signals (antenna or microphone) and with 2 different oscillator frequencies: about 24 MHz on receive mode or 10.695 +/- 1.5 kHz depending on sideband used when on transmit. The resulting interfrequency signal is fed into an SSB filter which is terminated by 2 resistors of 1k each side to ensure proper impedance matching.
Next stage is the interfrequency amplifier which is the same circuit like the receiver’s preamplifier. This one is also connected to the AGC’s DC line. On transmit mode with no AF signal on the AGC present this stage runs on full gain.
The chain is completed by the second mixer serving as product detector on receive and as tx mixer when going “on the air”.
The audio and AGC section
This section also is no “rocket science”. A simple preamplifier using a bipolar transistor (BC846) in common emitter mode, a low pass filter (R=1k and C=0.22uF) and an LM 386 amplify the resulting af sig to an adequate volume to listen to the sound even in an environment that is not 100% noise free.
The RF power module
This is a circuit I have built several times and it’s capable of delivering up to 5 watts of rf power. In this transmitter I’m not driving it beyond 3 watts which is suffice to establish connections on the 20 meter band worldwide a well performing antenna provided.
Emitter degeneration and negative feedback are present in preamp and driver stage to ensure maximum linearity. Both stages are operated in class A mode. The final stage works as a push-pull stage using class AB. Push-pull mode eliminates even order harmonics by circuit feature. A heatsink is mandatory for the final stage (the mounting frame in my case) and at least recommended for the driver. The values for the broadband transformers are stated in the schematic above.
The VFO module
This one is equipped with the clock oscillator chip Si5351A by Silicon Labs. I use it mounted to the well known Adafruit breakout board that can handle 5 volts even if the chip is designed for 3.3V. So, this board is compatible to standard microcontrollers like the ATmega168 that is applied in my transceiver. The display is the 1306 chipset based OLED that is also designed for 5 volts supply voltage. Both, the Si5351 and the OLED are designed for I²C-interface which is called “Two Wire Interface” (TWI) in Atmel’s language. The major advantage this interface has got is that only 2 control lines are required, one of them clock (SCL) and the other data line (SDA) to transfer data to the respective units. Basically you need two pull up resistors to tie these lines to +5VDD but I use the internal pull-up resistors in the Atmega168’s ports that do the job well.
Problem to be mentioned: When testing the early version of the receiver I found the OLED to be very noisy. After a brief research I realized that the signals that were audible in the receiver traveled on the VDD line. Thus I inserted an 82R series resistor and a set of blocking capacitors in the place which made the noise fully disappear.
The main board of the transceiver is made of a 5 by 7 cm breadboard with double-sided soldering pads each connected by a small tubing electrically connecting the both sides of the pad . This is a big advantage when you solder SMD components because you can setup the circuit on both sides of the board and save a lot of space. NExt is that it is nearly impossible to dissolder the pads even if you are resoldering the spots many times. The reason: The soldering pads are rivets anchored on both sides of the carrier of the board.
The components that aren’t available in SMT are standard through-hole but there are only a few like the SSB filter or some old 40673s I used instead of e. g. BF991. Coils for low power are wound on TOKO 5.5 mm coil formers. Power rf transformers are connected to soldering nails in the board. The inside view plus a centimeter scale:
On the left you can see the front panel with the controls, behind the panel the OLED and then the main rf board. The Si5351 is mounted vertically on top left from the AGC board. Underneath there is the receiver’s front end (hidden by the red power supply cable). On the right I sited the power transmitter, at the bottom there is the relay for tx/rx switching. All is built into a 9 * 5.5 * 4 cm Aluminum frame.
Being “on air” with the rig
Operating is really fun with this micro transceiver. Since the finishing of the transceiver 4 weeks ago I was on air daily. Here are some regions of the world I could establish successful contacts with. Antenna is a Delta Loop fed in on upper corner about 12 meters above ground.
Plans for the future
Several options may be my next project: First I got plans are to expand this rig to a multibander (due to Si5351’s capabilites of generating max. 160MHz signals), build a PA of about 30 watts max. power (with 2 transistors 2SC1969 in push-pull mode) or to rebuild the transceiver for 17 meter band and hoping that conditions will be better the next couple of years. Let’s see what the real deal will be! 😉
The demodulator section of the transceiver’s receiver starts with the product detector, which is made of another SA602. To get more audio volume a preamplifier has been added before the LM386 follows.
The AGC section hast got 2 crucial components: One resistor (this case 100k) and an electrolytic capacitor (in this case 100uF): They determine the time ramp for the AGC regulation curve. This means they define the response and decay time for the AGC and thus should be made easily changable for example by putting them into socket strips.
Hint: In certain cases it can be useful to add a potentiometer to give you control on the audio input of the AGC preamplifier.